Optimization approach to guaranteed state estimation of dynamic systems (Q5942031)

In previous papers of the author, a close connection between algorithms of optimal point estimation (e.g. the least-squares method) and set-theoretic guaranteed estimates is established as applied to problems of dynamic system state estimation. Here these studies are generalized to a wider class of observed dynamic processes. For the purpose of solving the guaranteed-estimation problem, a non-stochastic analog of the Bayes formula is applied. The basic result of the paper is that a Bellman-type functional equation indicating a constructive path of solving the problem is put into correspondence with the Bayesian set-theoretic algorithm.